By definition, systematic reviews answer specific, concrete questions. If you begin searching without defining your question, your search will take longer and be more difficult. You will also risk bias because your ideas might be influenced by what you find.
PICO is a common way to define research questions.
P: Patient or Population being studied
I: Intervention or Exposure affecting the population being studied
C: Comparison ie., to other possible interventions
O: Outcome = how you're measuring the success of the intervention
Depending on your question, you might also include a fifth line on Method/Study, in which you would specify the type of study you will look at.
EXAMPLE: Is poison-baiting effective at eradicating rats from islands?
Patient or Population: rat populations on islands
Intervention or Exposure: poison-baiting
Comparison: no baiting (or other methods)
Outcome: eradication of rat populations
Notice how this structure forces the researcher to be more specific. With an example like this one, you would also want to clearly define each term.
Some people find brainstorming techniques such as concept mapping to be helpful. Concept maps are linked nodes of concepts or ideas; the linking arrows demonstrate connections between nodes.
Source: “concept_map_lrg,” jean-louis zimmermann 
There are tons of different software options available to assist with concept mapping.
Here are a few browser-based tools to get you started:
A systematic review aims to be as unbiased as possible. One kind of bias is in how you conduct your search.
Students often make two mistakes, by searching only for:
Avoid these types of bias:
You will also need to assess the bias of studies you find. This is discussed in Critical Appraisal.
Source: "Studying," Skakerman
Before searching, you will need to know what you're searching for. It helps to have specific inclusion and exclusion criteria. Some possible criteria:
